Handling uncertainties in image mining for remote sensing studies

Abstract

This paper presents an overview of uncertainty handling in remote sensing studies. It takes an image-mining perspective and identifies different ways of handling uncertainties. It starts with the pixel, and through object identification and modelling, proceeds towards monitoring and decision making. Methods presented originate both from probability- and fuzzy-logic-based approaches. The paper is illustrated with three examples, one from a geographic information system stored object, one from an object identified from a remotely sensed image directly and a practical case study from the Tibet plateau. An important remaining topic is to combine and integrate errors and uncertainties collected during the whole image-mining process.